THE IMPACT OF PRODUCT INNOVATION ON

FIRM GROWTH USING A MULTI-STAGE MODEL: EVIDENCE IN A PERIOD OF ECONOMIC CRISIS

Pablo Garrido Prada Desiderio Romero Jordán

Maria Jesús Delgado Rodriguez

FUNDACIÓN DE LAS CAJAS DE AHORROS DOCUMENTO DE TRABAJO

Nº 796/2018

De conformidad con la base quinta de la convocatoria del Programa

de Estímulo a la Investigación, este trabajo ha sido sometido a eva-

luación externa anónima de especialistas cualificados a fin de con-

trastar su nivel técnico. ISSN: 1988-8767 La serie DOCUMENTOS DE TRABAJO incluye avances y resultados de investigaciones dentro de los pro-

gramas de la Fundación de las Cajas de Ahorros.

Las opiniones son responsabilidad de los autores.

1

The impact of product innovation on firm growth using a multi-stage model: Evidence in a period of economic crisis

Pablo Garrido Prada*

Desiderio Romero Jordán** Maria Jesús Delgado Rodriguez**

ABSTRACT

The period of economic crisis started in late 2008 meant a change in the conditions

of the environment, reducing the number of innovative companies as well as

investment in innovation. An adverse macroeconomic environment can also

determine the outputs of innovation and its effect on firm growth. The objective of the

article is therefore (i) to analyze the determinants of innovation output and (ii) to

examine how these outputs influences firm growth during a period of economic crisis.

For this purpose, we used a panel of Spanish firms during the period 2005-2013,

employing a sequential multi stage approach based on four phases: decision to

innovate, how much innovate, innovation outputs and finally, the effect of these

innovations on firm growth. The results confirm that for the period of economic crisis,

investment in innovation remains decisive for generating innovations outputs, albeit

with a slightly smaller effect. In contractive periods, both previous experience and

continuous activities of intramural R&D are two fundamental factors to generate

outputs from innovations. In any case, the positive effect of innovation outputs on firm

growth is reduced by half in times of crisis. The results imply that companies must

recalculate the opportunity cost of innovation, given that both previous experience in

innovation and a continued effort in R&D can reduce the negative effects of

contractionary macroeconomic periods on innovation.

Keywords: Product innovation, economic crisis, firm growth, multi-stage model,

Spanish firms.

Corresponding author: Pablo Garrido Prada. E-mail: gapra.pablo@gmail.com

* Facultad de Ciencias Económicas y Empresariales; Departamento de Economía Aplicada, Universidad Autónoma de Madrid, C/ Francisco Tomás y Valiente, 5, 28049 Madrid, España.

** Facultad de Ciencias Jurídicas y Sociales; Economía de la Empresa (ADO) Economía Aplicada II y Fundamentos de Análisis Económico, Universidad Rey Juan Carlos, Paseo de los Artilleros s/n, 28032, Vicálvaro, Madrid, España.

2

1. INTRODUCTION

The relationship between innovation and firm growth (and productivity) has been

widely studied for many decades, both at the macroeconomic and microeconomic

levels (Audretsch, Coad, & Segarra, 2014). There is considerable support that

innovation (leading to technological progress) is one of the main driving forces of

economic growth (Romer, 1990), in which firms play a core role to introduce those

innovations in the society and markets. However, the results about the relationship

between innovation and firm growth is usually obtained, and theoretically discussed,

on the background of period of macroeconomic expansion (Antonioli, Bianchi,

Mazzanti, Montresor, & Pini, 2013).

The global financial crisis, started in late 2008 hit many countries in the world,

devastating their economies, nearly collapsing the banking systems and decimating

the financial resources of their companies and citizens in their countries (Hausman &

Johnston, 2014). Previous studies found a sharp reduction in the number of

innovative firms and in innovative investment during the last period of crisis for many

develop countries (Archibugi, Filippetti, & Frenz, 2013b; Filippetti & Archibugi, 2011;

OECD, 2009, 2012; Paunov, 2012). However, this study aims is to give a step further

and analyse the determinant and consequences of innovation on firm growth in a

period of economic crisis. What is the role of investment in innovation over innovation

output during downturns? Does the outputs of innovation have the same effect on

firm growth in periods of crisis as in expansive ones? Which other factors determine

the innovation output and firm growth in period of crisis?

The severity and pervasiveness of the economic recession stimulated the analysis of

the relation between innovation and firm growth on the crisis challenges. (Antonioli et

al., 2013). Develop successful innovations is a difficult task, which depend among

other, on firm and market characteristics, time period and macroeconomic

environment (Audretsch, Segarra, & Teruel, 2014). Li and Atuahene-Gima (2001)

proved that the innovation effectiveness is to large extent determined by

environmental factors and institutional support. Therefore, it is to be expected that

during downturns, the challenge to innovate and gather benefit from them, is likely

amplified given demand uncertainty, revenues decline, and financial constraints

(Amore, 2015).

3

For this purpose, we used a panel of Spanish firms from 2005-2013. Spain entered

into a deep recession in 2008, due to lack of liquidity, rising defaults and debt that

caused a bank bailout, 25,5% unemployment rate in late 2013 and long GDP

constrictions from 2008 to 20131. Thus, inspect Spanish firms provides an

appropriate setting for an analysis of innovation in a period of economic crisis. We

estimated a sequential model (generally called CDM models) based on four stages:

innovation decision, innovation investment, the innovation outputs and finally the

effect of innovation output on firm growth.

The paper is structured as follows. Next Section 2 reviews the literature. Section 3

describes the data and the sample selection criteria. Section 4 defines the model, the

estimation method and variable. The results are presented in section 5; and we finish

with conclusions and a discussion of the research in Section 6.

2. LITERATURE REVIEW

2.1 Innovation output determinants

There is a large volume of published studies analyzing the impact of innovation on

growth and company productivity since the work of Griliches (1986). The study of the

effect of innovation on firm performance has evolved, from a perspective of the inputs

of innovation to an output perspective (Crepon, Duguet, & Mairessec, 1998). The

underlying idea is that investing in innovation is a necessary but not sufficient

requirement for introducing successful new products, services, processes,

organization and marketing methods for companies (Bessler & Bittelmeyer, 2008).

Only the innovations implemented are those that have the capacity to modify the

performance and growth of the firm.

Thus, innovation output means the company's ability to introduce new or significantly

improved products, services, processes, methods of organization and distribution for

the company or market (Oslo Manual, 2005).

According to the Shumpeterian theory, innovation allows to overcome the

equilibrium, gaining profits through a more dominant position in the market (Arrow,

1962). The innovation expands not only the limits of knowledge, but also the market,

replacing the current technology. Thus, through investment in innovation, the 1 Source for the Spanish GDP: AMECO.

4

company can generate outputs of innovation as new products, processes, methods

of organization or distribution. This helps to produce at lower cost and open new

markets that end up having a positive impact on the productivity and growth of the

firm (Bayus, Erickson, & Jacobson, 2003). The accumulation of technology, both

tangible and intangible, is difficult to replicate by the firm’s competitors in the short

term (Freeman, 1994), and it creates new opportunities that non-innovative firms

cannot achieve (Breschi, Malerba, & Orsenigo, 2000).

Hence, it is expected that the greater innovation input (investment), the greater

outputs of innovation. However, not many studies have found this positive

relationship (Klomp & Van Leeuwen, 2001). This is because the innovation

performance depends not only on investment in innovation, but also on the objectives

of that innovation, the type of innovation, the sources of information, the cooperation,

the public aid, the sector to which it belongs, the internal characteristics of the

company itself, and on the macroeconomic environment. The sum of investment in

innovation plus all factors determine the ability to introduce new products, services,

processes, organization and distribution (Hashi & Stojčić, 2013).

As any other investment, innovation is not risk free. In times of economic crisis,

uncertainty about the macroeconomic environment coupled with the complexity of

technology may undermine and underestimate the potential benefits of innovation

(Miller, 2006). Firms (and financial markets) are more reluctant to invest in

innovations when the risks of such innovation are higher because of uncertainty. This

may lead companies to simply defend their position in the market (Auh & Menguc,

2005) and prefer to exploit existing technology and knowledge (March, 1991). If the

company does not have a clear commitment to innovation, new knowledge may not

be enough to radically challenge the existing knowledge base and the technology

driving sales (Choi & Williams, 2014). This implies that the new knowledge generated

is not sufficient to sustain sales growth, or to taking advantage of the economies of

scale of learning that stem from innovations. In this way, investment in innovation

may not increase the outputs of innovation in times of crisis in comparison with the

expansion periods.

5

2.2 Innovation output and firm’s growth

Companies innovate seeking the outputs of the innovation that may lead to a

competitive advantage. New products or services imply new markets, which can

increase sales (Bayus et al., 2003). Innovation output can also improve productivity

and growth by increasing the ability to use the company's assets, and by promoting

technological changes. In addition, Innovations develop and update routines, which

favors the adaptability and use of company resources. These routines help the

spread of new knowledge to any company activity increasing economies of scale

(Choi & Williams, 2014). Although the innovation of a new product, service -or

process-, cannibalizes old products, services and manufacturing methods, which

may reduce productivity and growth. It is also an evolutionary method that generate

sustainable competitive advantage against its competitors, and create market power

through sales increases, and cost reduction.

Previous studies have found a positive relationship between innovation output and

firm growth, both in sales (e.g. Coad & Rao, 2010) and in employment (e.g. Dachs &

Peters, 2014; Harrison, Jaumandreu, Mairesse, & Peters, 2014). There is also an

extensive literature that finds a positive relationship between innovation output and

productivity (see B. H. Hall, Mairesse, & Mohnen, 2010; Mohnen & Hall, 2013).

Both, the Shumpeterian and the evolutionist theory advocate innovation as the

engine to overcome the economic crisis. For the Shumpeterian, radical innovations

leads the company to give a qualitative leap that generates new demand, reduction

cost, extra profits and investment. These radical innovations affect positively the

macroeconomic as well as help to overcome the constraints that occur in

contractionary periods. In the same way, evolutionists suggest that innovations allow

the updating of technologies, routines and processes, which improve the benefit of

the company. Only the best adapted survive and allow the economy to evolve to

overcome the economic crisis.

Apart from the internal factor, external environment can moderate the effect of

innovation output on firm growth. In periods of crisis, the company has to face higher

levels of uncertainty, lack of financing and lower demand, which can break the

positive relationship between innovation performance and growth. From the supply

perspective, the economic crisis can lead to invest inefficiently in innovation, due to

6

lack of resources or information. This can undermine the ability to introduce radical

innovations in both the company and the market. If these innovations are not

differentiated enough, it can reduce the potential performance of innovation, which

directly affects the sales of that good or service. From the perspective of demand, in

times of crisis, when consumer disposable income is lower, preferences vary

focusing on covering basic needs at the lowest possible price. This can affect sales

of innovative products, which usually have a higher manufacturing cost in the early

stages of the product. Altogether, the benefits of product innovation can be affected

in times of crisis. As discussed earlier, investing in innovation is not synonymous of

obtaining innovation output in crisis, and likewise, being able to produce innovation

output may not ensure an increase in sales, employment or productivity of the firm

before under macroeconomic conditions.

3. THE DATABASE

The Technological Innovation Panel (PITEC) is a panel survey to study the

innovation activity of Spanish firms over time. PITEC consists of repeated

observations on the same firms over time, with annual statistical information on their

innovation activities. It follows the guidelines in Oslo Manual (OECD, 2005) and

Frascati Manual (OECD, 2015) using a standardized questionnaire. The database is

based on the Spanish Innovation Survey and the Community Innovation Survey

(CIS). PITEC is carried out by the INE (The National Statistics Institute), which counts

on advice from a group of university researchers and the sponsorship of FECYT and

Cotec2.

Specifically, PITEC provides detailed information on the innovation activities of

Spanish firms for the period 2003-2014. For instance, it offers information on different

types of innovation, R&D expenditures, R&D employees characteristic, innovation

outputs, cooperation, or funding to undertake innovation activities. For reasons of

opportunity and viability, PITEC started with two samples with data from 2003: a

sample of firms with 200 or more employees (sample of large firms, which

represented 73% of all firms with 200 or more employees according to data from the 2 FECYT: Spanish Foundation for Science and Technology. COTEC: Spanish Foundation for innovation. PITEC web page: https://icono.fecyt.es/PITEC/Paginas/por_que.aspx DIRCE: Central Directory of Companies in Spain.

7

DIRCE), and a sample of firms with intramural R&D expenditures. Given the

improvements made by the INE in information on firms undertaking R&D activities,

there were enlargements of the second sample in 2004 and 2005. Thus, in 2004, the

database was expanded with a sample of firms with fewer than 200 employees,

external R&D expenditure and no intramural R&D expenditure; and in 2005 with a

representative sample of firms with fewer than 200 employees and no innovation

expenditure. This database has been used for several articles (e.g. Coad, Segarra, &

Teruel, 2016; Segarra & Teruel, 2014).

Our sample contains information for the period 2005-2013, since information in 2003

and 2004 is limited. In the pre-crisis period, the annual GDP growth rate was 3.5%.

The financial crisis began in 2008, moving from a Spanish GDP growth of 1.02% in

the first quarter of such year to a decline of 1.73% in the first quarter of the following

year. Although economic growth remained constant (0.2%) in 2010, it started to

decline again in 2011 until the end of 2013 with an annual GDP growth rate of -1.7%.

In 2013 was the last year of the Spanish economic recession.

The initial sample contains 8492 firms (76428 observations) belonging to all sector.

We eliminate observations that included some kind of incident (problems of

confidentiality, sales or employees variation due to takeovers, mergers, or

bankruptcy, etc.) and those with an obvious anomaly variable (such as null or below

to 10000 euros sales). However, we did not drop outliers’ observation that had been

winsorized to avoid estimation bias. Finally, we keep those firms that remain with

information in the whole period 2005-2013. Thus, the final sample consisted of a

balanced panel of 6661 firms (59,949 observations). Our final sample include firms

from all sector.

Table 1 and 2 summarize some descriptive statistics about innovation. Table 3

inspects the average and mean firm’s sale growth (in percentage) in different groups

of firms. Starting by table 1, we find that the proportion of innovative firms, follow

Manual Oslo (2005) definition, decreased during the period of crisis especially for

technological innovations (product and process). Since the definition of innovative

firm refers to the period t to t-2, in year 2011, there was a clear decline in the

proportion of innovative firms. This year includes innovative companies in the period

2009, to 2011. The decline was keeping up until the end of the crisis. It also

8

highlights that the decline was stronger in technological innovations (product or

process) which is usually associated with greater expenses.

Nevertheless, the incidence of the crisis was uneven depending on the size of the

company. In micro and small enterprises, the proportion of firms that stopped their

innovation process was larger than in medium and large companies.

Insert table 1 and 2

In this way, the size of the company has been fundamental for the decision and

capacity of the company to embark on innovative activities. Is innovation spending

also contingent on the size of the company? As can be seen in table 2, innovation

spending on innovative firms also declined during the economic crisis, being more

marked in micro and small firms. Yet, spending on medium and large companies,

although slightly lower, remained. This may mean that innovative firms, especially

larger ones, keep their spending on innovation even in times of economic crisis. One

of the items of expenditure that represents greater outlay for the companies is the

acquisition of machines, equipment and software for innovation. Again, we found that

large companies expend more in this item than smaller ones. However, the

decreasing during the crisis was similar and not clearly associated with company

size. Finally, the percentage of sales of new products for the company or the market

due to product innovation also remained constant throughout the period, except for

micro and small firms in which there was a slighting tendency to a decrease.

Analysing our second variable of interest in Table 3, we see that innovative firms had

a slightly higher level of firm’s sale growth than non-innovative ones. These

differences remained constant for the period analysed. Likewise, companies with

export capacity maintained a higher level of firm’s sales growth during the period of

crisis –not in macroeconomic expansion period- compared to the rest for the whole

period analysed. Focusing on size, figure 1 portrays the sales variation by size and

year. In year 2008, there was a big decline in the firm growth (but still positive)

coinciding with the beginning of the crisis in late 2008. The firm´s growth turned to

negative in 2009 and kept lower level of variation since them.

Insert table 3

Insert Figure 1

9

4. MODEL SPECIFICATION

The analysis in this paper is based on a structural model (often called CDM)

consisting of four stages (Crepon et al., 1998): the decision of firms to innovate; their

decision about the amount of innovation expenditure; the production of innovation

output; and, the impact of innovation output on productivity (or growth). The model is

calculated sequentially, assuming causality between the decision to innovate and the

firm growth, but also allowing the inverse causality. Each of these stages includes

determinant factors, such as firm’s characteristics as well as, environmental and

institutional characteristics. The first two stages estimations –decision and

investment in innovation- are reported in appendix A for space limitations.

The use of a panel of data allows us to solve some of the limitations of previous

studies based on CIS. On the one hand, the innovation is carried out over several

years and different costs are allocated depending on the development phase of the

innovation. We correct for this fact imputing the average expenditure on innovation

by firm in period t to t-2. On the other hand, the output of innovation is not immediate,

so having a data panel allows us to introduce delays in the explanatory variables by

increasing the robustness of our model. Finally, with this panel from 2005 to 2013.

Similarly to the studies such as Hashi and Stojčić (2013) and (Lööf & Heshmati,

2002, 2006), the decision to innovate and the decision on how much to invest in

innovation are linked to their determinants in the first two stages of the innovation

process. Then, the third stage is a knowledge production function linking innovation

input and output. Thus, the innovation output, included in the third equation depend

on the decision to innovate (first) and how much to invest in innovation (second

equation). Finally, in the fourth stage the productivity of a firm is related to the

innovation output.

The literature on innovation and firm performance identifies two major problems with

the econometric specification of this relationship, namely simultaneity and selectivity

biases (Hashi & Stojčić, 2013; Mairesse & Robin, 2017). The first one arises because

some factors influence in chorus on firm’s decision to innovate, how much spent, and

its final performance. Selectivity bias arises from the fact that not all firms are

engaged in innovation and some innovations are not successful.

10

This sequential structural model approach includes the predicted value of one

endogenous variable that enters in the estimation of the next equation. The inclusion

of a correction factor for potential selection bias relaxes the original CDM model

assumption that errors should not be correlated.

4.1 General Specification of the Model

Similar to the CDM, we use a four-equation structural model in our sensitivity

analysis:

Let i=1,…N index firms, and t=1…9 index year time from 2005 to 2013. The first

equation account for firms’ decision to innovate in the period t to t-2; the second one

account for firms’ innovate effort in the period t to t-2. Being 𝑑𝑖(𝑡;𝑡−2)∗ and 𝑔𝑖(𝑡;𝑡−2)

∗ two

unobserved (latent) variables of the decision to innovate and of the level of the firm’s

investment in innovation respectively, we can define:

𝑑𝑖(𝑡;𝑡−2) = 𝛽0𝑥𝑖(𝑡;𝑡−2)0 + 𝑢𝑖(𝑡;𝑡−2)

0 (1)

𝑑𝑖(𝑡;𝑡−2) = 1 if 𝑑𝑖(𝑡;𝑡−2)∗ > 0 , otherwise 𝑑𝑖(𝑡;𝑡−2) = 0

And �𝑔𝑖(𝑡;𝑡−2)� 𝑑𝑖(𝑡;𝑡−2) = 𝛽1𝑥𝑖(𝑡;𝑡−2)

1 + 𝑢𝑖(𝑡;𝑡−2)1 (2)

𝑔𝑖(𝑡;𝑡−2) = 𝑔𝑖(𝑡;𝑡−2) if 𝑔𝑖(𝑡;𝑡−2)∗ > 0 , otherwise 𝑔𝑖(𝑡;𝑡−2) = 0

Where 𝑑𝑖(𝑡;𝑡−2) is an observable variable of firm’s decision to innovate in t to t-2 and

𝑔𝑖(𝑡;𝑡−2) is the observable level of the average firm’s investment in innovation in the

period t to t-2. Then, 𝑥𝑖𝑡0 , 𝑥𝑖𝑡1 , 𝛽0, 𝛽1 are independent variables and their corresponding

parameters which reflect the impact of different determinants on the firm’s decision to

invest in innovation and on the level of expenditure on innovation. 𝑢𝑖𝑡0 , 𝑢𝑖𝑡1 are random

error terms with zero mean, constant variances and not correlated with the

explanatory variables.

The third equation are the innovation production function (innovation output)

presented as follow:

𝑘𝑖𝑡 = 𝑎𝑔2𝑔�𝑖(𝑡;𝑡−2)+𝛽2𝑥𝑖𝑡∗2 + 𝑢𝑖𝑡2 (3)

Where 𝑘𝑖𝑡 represents the innovation output of firm in year t, 𝑔�𝑖(𝑡;𝑡−2) represents

estimates of innovation input from Equation (2) and 𝑎𝑔2 is the corresponding vector of

unknown parameters. 𝑥𝑖𝑡2 is the vector of other explanatory variables which includes

among others inverse Mill’s ratio estimates from Eq. (1) and performance from the

11

fourth stage to control for selection bias and feedback effect. Finally, 𝛽2, are the

coefficients of the explanatory variables while 𝑢𝑖𝑡2 is the random error term with mean

zero and constant variance not correlated with explanatory variables. By using its

predicted value, we also instrument the innovative effort 𝑑𝑖(𝑡;𝑡−2)∗ and take care that it

is possibly endogenous to the innovation production function (Griffith et al, 2006).

Finally, the last equation of the model relates the innovation output and other factors

with the firm’s performance (growth). We used a log transformed function similar than

a Cobb-Douglas production function as follow:

𝑞𝑖𝑡 = 𝛼𝑘�𝑖𝑡 + 𝛽3𝑥𝑖𝑡∗3 + 𝑢𝑖𝑡3 (4)

with 𝑞𝑖𝑡 as the dependent variable indicating the firm’s rate of growth in year t; 𝑘�𝑖𝑡

representing estimates of innovation output from Eq. (3), where 𝛼 the elasticity of

production with respect to changes in innovation output. The vector 𝑥𝑖𝑡∗3 is a vector of

input variables with include among other one lag of the dependent variable, 𝛽3 is the

correspondent unknown parameters and 𝑢𝑖𝑡3 being the error term which is assumed to

be uncorrelated with explanatory variables.

4.2 Econometric technique

As we said before, the first two equations are estimated jointly through a double

hurdle model (Cragg, 1971) in which observations on both innovative and non-

innovative firms are included. This methodology separates the decision to innovate in

two process. The first hurdle corresponds to factors affecting the decision to innovate

(participation) and the second to the decision of level of expenditure. A different

latent variable is used to model each decision process. This model is closer to the

firm reality where some companies do not want to innovate, and some other, do not

have the enough resources and capabilities to do it. Thus, the first hurdle needs to be

crossed to be a potential innovator. Given that the firm is a potential innovator, it

current circumstances dictate whether innovate—this is the second hurdle (Engel &

Moffatt, 2014).

Two main advantage emerges using a double hurdle model: First, the probability of a

positive value (first stage) and the actual value, given that it is positive (second

stage), can be determined by different underlying process (i.e., different parameters)

overcoming the limitations of the Tobit model (Burke, 2009; Engel & Moffatt, 2014).

12

Second, the double hurdle model can be interpreted as a flexible version of the

Heckman model. In the Heckman model, zero observations arise due to

nonparticipation solely whereas the double hurdle model relaxes this assumption and

allows zero observations to arise in both the participation hurdle and expenditure

hurdle. (Eakins, 2016). Therefore, the double hurdle model features both the

selection mechanism of the Heckman model (which is not a feature of the Tobit

model) and the censoring mechanism of the Tobit model -which is not a feature of

the Heckman model- (Eakins, 2016).

However, is reasonable to think that the decision to innovate (participation) and how

much to spent (expenditure) are related decisions, but it may be driven by different

variables. The Double Hurdle model assumes that there is no correlation between

the error terms in the two hurdles (Cragg, 1971), so we relaxed this assumption

included the inverse Mills ratio from the first component as explanatory for the

second component (Burke, 2009; Heckman, 1979; Wooldridge, 2010). Thus, we

assumed that there is a correlation between the decision to innovate (participation),

and how much to spend in innovation (expenditure). Finally, we correct for

autocorrelation using cluster- robust standard error.

The third and fourth stages are estimated as a system in a 3SLS simultaneous

equation where the endogenous innovation output variable is limited only to strictly

positive values in the last step. In the production function model illustrated by

equations (2) to (4), the innovation input g in equation (2) is an explanatory variable

in the innovation output equation (3), and innovation output, k, is an explanatory

variable in the productivity equation (4). Because of the endogeneity of these

variables, we cannot assume that the explanatory variables and the disturbances are

uncorrelated. Thus, similarly than Lööf and Heshmati (2002, 2006), van Van

Leeuwen and Klomp (2006) and Hashi and Stojčić (2013), we relaxed the full

correlation between the error terms of the fourth equations with one of partial

correlation between disturbance terms. It is assumed that the disturbance terms from

the first two stages of the innovation process are correlated with each other because

of unobservable characteristics of firms. As noted by Griffith, Huergo, Mairesse, and

Peters (2006) such procedure controls for the possibility of potential endogeneity of

innovation input to the innovation output. Similarly, the third and fourth stages are

estimated jointly as a system in a reduced form of the model.

13

4.3 Definition of variables

In this subsection, we describe the variables used for each stage of the model.

Information about how is measured each of the variables is available in Appendix A.

Decision to innovate

A firm decision to innovate is considered (dependent variable in Equation 1) as long

as it reported a positive value for innovation expenditure in the period t to t-2. Similar

that Hashi and Stojčić (2013) or Peters, Roberts, and Vuong (2017), I broader the

innovation definition included R&D expenditure and expenditure on machinery,

equipment, software, patents, know-how and training of staff for innovation activities

(Oslo Manual, 2005). Most studies have adopted the practice of including the same

variables as determinants of the decision to innovate and how much invest in

innovation. However, in the explanatory variables of Equation (1) we included only

variable available for the whole sample (innovative and not innovative firms) to avoid

spurious correlations. Variables such us organizational and marketing innovations,

objectives of innovations should be avoided since this variables are only answers by

innovative firms.

Among the factors that are usually included in the decision to innovate as well as

how much to spend on innovation are the size of the company, the export capacity,

types of cooperation, source of financing, company structure, sector and previous

experience.

We divided the exploratory variables of the decision to innovate in firms’

characteristics, and market characteristics. Firms characteristics includes, the size of

firm, belong to a group, foreign ownership participation, firm age, capacity to invest,

internationalization, experience in innovation and firm localization in a technological

park.

Firm size is measured by logarithms of firm’s employees. Two dummy variable

identified if the firm belong to a group of enterprises being the parent or subsidiary.

Then a dummy variable if firm is private and with foreign participation. The firm age is

measured in logarithm, whereas a categorical variable of the firm degree of

internationalization (abroad UE border) in the year t-2 identified export capacity. The

firm capacity of investment is measured by the logarithm of the firm average gross

investment in tangible assets in the period t to t-2. Innovation experience is imputed if

14

the firm was able to introduce a successful innovation previously (in t-3), or in the first

year that the firm was included in the sample. Finally, a dummy identified if the firm is

established in a technological park.

In relation with the market characteristics, we included factors hampering innovation

(cost, knowledge and market); Spanish representative sector dummies; and a

dummy variable of the economic crisis period started in 2009 (in the late 2008).

Innovation investment

In the second stage, Innovation input was defined as the natural logarithm of the

average amount spent per year in the period t to t-2 on innovations divided by the

average employees per year in the same period. This definition measures the

innovation expenditure intensity and it is widely used in the field. However, imputed

the value of three years represents a huge advantage with respect other research.

One of the limitations of the analysis based on CIS databases is that it can only

allocate the expenditure on innovation of year t. Yet, just as the process of innovation

takes place over several years, innovation expenditures are also imputed over more

than one period or year. In many cases, innovation expenses are paid in several

years, (e.g. through a financial agreement), or the innovation process has different

phases that require diverse levels of expenditure. Impute only the year t may bias the

innovation input data and its effect on innovation output (Archibugi, Filippetti, &

Frenz, 2013a). Further, defined this way, the variable encompasses spending on all

innovation activities mentioned earlier (intramural and extramural R&D expenditure,

investment in machinery, equipment and software and other acquisitions of external

knowledge).

The exploratory variables of Equation (2) includes the same exploratory variables of

Equation (1) except the dummy variable if the firm was able to introduce a successful

innovation previously. Further, we also included variables that may affect the

investment in innovation. as dummies identifying highly important sources of

information about innovation (internal sources, market sources, institutional sources);

firm access to subsidies thought a dummy if the firm receives any public financial

support from national or EU institutions; a dummy variable if the firm is involved in

continuous activities of intramural R&D; a dummy variable if the firm invest in

external R&D; breakdown of expenditure on innovation as percentage of total

15

innovation expenditure in the period t to t-2 on acquisition of machines, equipment

and software; three dummies variable if the firm cooperate on innovations with: (1)

other abroad enterprises or institutions, (2) public institutions and research centre (3)

competitors and other firms in the same main market segment (coopetition) in the

period t to t-2.

Innovation output

In Equation (3) innovation output is measured by the natural logarithm of the share of

sales of new products and services (new to a firm and new to the firm’s market) of

the firm. As Oslo Manual (2005), we considered “new” as entirely new, or

substantially improved good of service. Thus, it is the percentage in logarithm of total

sales in year t due to products or services launched in the period t to t-2.

Unfortunately, for the other types of innovation – process, organizational and

marketing- the only innovation measures available are dichotomous measures. The

sale of new products is considered as the most robust measure since the introduction

of new products or services includes the whole process of innovation and allows

quantifying the commercial success of innovation (Mohnen & Hall, 2013).

The explanatory variables in this equation are also divided in firm and environment

characteristics. Firm characteristics included innovation investment intensity

measured by innovation input from the second stage; a dummy variable if the firm

was able to introduce a successful innovation previously (in t-3); a dummy variable if

the firm is involved in continuous activities of intramural R&D in the previous year; a

dummy variable if the firm invest in external R&D in the previous year; two dummies

if the firm was able to introduce process or non-technological innovation

(organizational and marketing) in the same period (t to t-2) respectively. We included

two dummies representing firms whose objective is explorative or accumulative

innovation (see Appendix A); three dummies for cooperation with abroad firm or

institutions, public institution, or with competitors respectively in the period t to t-2.

Further, two dummies if the firm received any public financial support from national or

EU institutions, etc. in the previous year. We also include the natural logarithm of firm

growth from the Equation (4); firm size; a proxy of the firm environment though a

dummy if the firm is established in a technological park; the percentage of employees

with high education; two dummies variable if the firm operates in high manufacture or

16

high services sector. Environment characteristics includes sector dummies and

economic crisis dummy.

Finally, to correct the sample selection of select only companies with positive values

of innovation output, we included two inverse Mills ratio: one from the first stage

controlling selection bias of the decision to innovate, and another one controlling for

the fact that firms may introduce other types of innovations that do not change the

dependent variable of the Equation (3). In other words, companies can innovate with

goals far from the introduction of new products for the market or the company. For

example, innovations aimed at adapting to new regulations, those that seek to

improve the quality and welfare of employees, process innovations that reduce the

environmental impact of a particular product or process, non-technological

innovations3.

Firm growth

Finally, we defined the dependent variables of four stage in Equation (4) as firm

growth measured through the variation of turnover. Since this variable is in

logarithms, we built an index variable with base in 2005. Firm growth variable is

frequently used in the innovation-performance analysis mainly due to the

anonymization that innovation surveys have. They do not include accounting

variables and the company name is not display (e.g. Audretsch, Segarra, et al., 2014;

Raymond, Mohnen, Palm, & Loeff, 2010).

As mentioned above, the equation (4) includes the predicted values of innovation

output from the previous stage, one year lag of the dependent variables which was

instrumented with a two year lag of the same variable. The independent variables

included are firm size; firm age (in logarithm); a proxy of the firm capacity of

investment, firm internationalization, foreign ownership participation, other types of

innovations, as well as sector and a dummy for the period of economic crisis. Firm

size and firm age is calculated in the same way as described before. We calculated

the firm capacity of investment by the logarithm of the firm average gross investment

in tangible assets in the period t to t-2. Firm internationalization is measured by the

3 We run an additional regression where the dependent variable is a dummy that takes the value 1 if a firm successfully introduced process or non-technological innovations in the period t to t-2 but not product innovation. The exploratory variables are the same ones introduced for equations 3 (innovation output).

17

percentage of overseas sales outside the UE. A dummy variable for being part of

group of enterprises and a dummy variable if firm is private and with foreign

participation is used for firm structure. We included a dummy if the firm was able to

introduced a non-technological innovation (organizational and marketing) in the same

period (t to t-2), We included We added four main sectors: trade, hospitality, services,

construction, and finance. Finally, we included a dummy for the period of economic

crisis started in 2009.

5. RESULTS

We present the results of the determinants of innovation output –third stage- and

the effect of innovation output on firm performance –fourth stage-. The determinants

of the decision to innovate and how much to invest in innovation are reported in

appendix C because space limitation. In the descriptive analysis in the previous

section, we found clear differences in depending on size and economic period. For

this reason, in the next tables, we included pre-crisis and crisis economic period, and

the differences between small (including micro) firms and large (including medium)

ones.

5.1 Innovation output

The third stage consists only on firms that have reported a positive amount of

innovation output measured by the share of sales coming from the introduction of

new product or services to the firm or market. Two inverse Mill’s ratio was included to

control for potential selectivity bias. One comes from the first equation and another

one for an auxiliary regression of the equation 3. The results of the estimation are

presented in table 4. The inverse Mill’s ratio from the first equation is insignificant but

not the inverse Mill’s ratio from the auxiliary regression suggesting the

appropriateness of correcting for selection bias.

Insert table 4

Starting by the determinants of the innovation output, we found differences between

pre crisis and crisis period. The coefficients of innovation input yields a positive

relationship with innovation output in the whole sample (0,244). Nevertheless, the

elasticity is higher in micro and small (SM) firms than in medium and large (ML).

There is also a decrease in the elasticity during period of economic crisis compared

18

with period of expansion. There are three likely explanation: One is that in period of

economic crisis, the objectives of innovation activities changes from product into

process or non-technological. Another explanation is that the demand change in

period of crisis, diluting the potential positive effect of new product or services.

However, as we could analyse in table 1, the average share of innovative products

and services launched remained similar in period of crisis. Thus, there are not clear

demand factors as changes in consumer’s preferences. Finally, it is also likely that

companies prefer to defend their position in the market exploiting the existing

technology and knowledge, which may lead to an inefficient investment in innovation

in period of crisis (Auh & Menguc, 2005; March, 1991).

In period of economic crisis, other factors become important to generate innovation

output. Firms involved in continuous R&D lead to an increase in the innovation

output. The coefficient is higher in period of economic crisis. Further, continuous

R&D generates higher innovation output in ML firms than in MS. This means that

those companies that have an established research department, for which they carry

out innovation activities on an ongoing basis, obtain greater innovation output than

those companies that innovate on an occasional basis. This is also related to what

we find in the previous section where ML seek exploratory innovations.

Furthermore, previous experience in innovation positively affects innovation output.

Again, the positive effect is higher in periods of economic crisis. Those companies

that had already introduced innovations in previous periods (t-3) have a greater

capacity to readapt to the new adverse macroeconomic conditions, and obtain

greater innovation performance in terms of introducing and sell new products or

services. This variable is not relevant for ML firms in period of expansion but

becomes significant in period of crisis. For MS firms, previous experience is essential

for successful innovation, especially in downturns. Again, the learning process and

experiences are relevant in period of downturn.

However, extramural R&D does not influence positively the innovation performance

in any kind of period. Therefore, the same conclusion arises: investment in innovation

is a necessary condition, but other key factor can enhances innovation output in

periods of crisis. This investment must be accompanied by previous experience and

knowledge.

19

Combine product and process innovation increase the innovation output. Both

strategies are related since the introduction of a new product and services usually

requires new processes. The coefficient of non-technological innovation yields the

same results for period of crisis than expansion. When the company is able to

innovate in several types of innovation increases the innovation output.

Further, those companies that have a clear objective of exploratory and cumulative

innovation increase the innovation output in period of economic crisis. The positive

effect of cumulative innovation is twice as much as exploratory innovation. This can

mean two things: First, is that because Spain is a "follower" country in terms of

technology, cumulative innovation causes a greater increase in the share of sales of

new goods and services. Second, is that in periods of crisis, changes in demand

preferences (e.g. more focus on price and basic needs) causes cumulative

innovations to perform better than the exploratory ones in terms of percentage of new

products or services in relation to their total sales.

The cooperation with other foreign institutions or firms influences the innovation

output positively in the whole period. The cooperation with abroad institution is

significant for MS firms but not for ML. This means that MS firms take advantages of

this type of cooperation to generate new product or services. Nevertheless,

cooperation with public institutions and competitors impact negatively in the

innovation output. The last result not imply that cooperation with public institution is

not effective, if not that cooperation with public institution may not have the objective

to introduce new product or services (e.g. focus in process or environmental issues).

Finally, the subsidies of Spanish public institutions allow, in times of crisis, to achieve

greater innovation output, especially for MS firms.

Analysing the firm’s internal characteristics, the effects of firm’s growth on innovation

output is positive in the whole period, without difference between MS and ML firms.

Firms with higher growth also gather higher innovation output in a feedback effect.

Finally, the environment is also contingent to the innovation performance. In period of

economic expansion, technology parks are positively related with innovation output.

Knowledge spillover in science parks produces a positive impact on firms that are

located together in terms of performance and efficiency (Audretsch & Feldman,

2004). Thus, firms are able to assimilate knowledge spill over for their neighbours.

20

However, this relationship disappear in period of economic crisis. One likely

explanation is that in period of crisis, the free share of information and knowledge is

smaller which can reduce the advantages of technology park.

5.2 Firm growth determinants

In this fourth stage, we analysed firm growth, which is determined by the

innovation output, as well as other control variables available in the database. Table

5 contains the estimation output by 3SLS.

Insert table 5

The results indicate that innovation output significantly increases firm growth. Again,

we find differences between the pre-crisis and crisis period and between micro and

small firms (MS) vs medium and large (ML). Whereas the elasticity of firm growth-

innovation output is 0.256 in period of expansion, it decreases to 0.112 in in period of

economic crisis. Thus, the effect of innovation output on firm growth is half in period

of economic downturns. There are also differences taking into account company size.

Whereas in MS firms, innovation output leads to increase firm growth, it is not the

case for ML firms where the positive effect disappears in period of crisis. Further, the

elasticity is higher in MS firms than in ML. For MS innovation is a clear strategy to

become a competitive firm.

Similarly, the introduction of process and not technological innovation does not

enhance firm’s growth in period of economic crisis. The non-significance of process

and non-technological innovation dummies is frequent once the intensity of product

innovation is introduced as a continuous variable (Lee et al., 2003; Criscuolo, 2009).

The coefficient of the lag of firm growth is statistically significant and positive,

indicating the persistence of the changes implemented by the company to improve its

productivity and growth. Firm size is positively related with firm growth. One likely

explanation is that the stronger benefits derived from dominance position, access to

internal and external financing of bigger firms leads then to increase their size.

Export and invest capacity increase firm growth, but its elasticity is relatively small

compared to the other variables. Export becomes relevant in period of economic

crisis. Those companies present in international markets can offset the decline in

21

domestic markets demand. Contrary, company age and be part of a group reduces

firm growth. Finally, the period of economic crisis reduces the firm growth.

6. DISCUSSION AND CONCLUSION

The economic crisis of 2008 changed the environment conditions, which affects the

decision to innovate and the total investment in innovation (e.g. Archibugi et al.,

2013b; Paunov, 2012). Similarly, the macroeconomic environment may also

determine the performance of these innovations and their final effect on firm

performance, and growth. In contractive environments, uncertainty, liquidity and

demand constraints may limit the benefits achieved from its innovation strategies.

The objective of the article has been (i) to analyze the determinants of innovation

performance and (ii) to examine how these outputs determine the company's growth,

during the expansionary and contractionary periods of the economy. With this aim,

we have obtained information from Spanish companies (one of the countries most

affected during the crisis of 2008) and we estimated a sequential model based on

four stages: decision to innovate, how much to innovate, the performance of

innovation and finally the effect of these innovation output in firm growth.

Effect of the economic crisis: The size of the company in the input and output of the

innovation.

This study corroborates that the crisis has reduced the number of companies that

decide to innovate in Spain, which decreases the total investment in innovation. This

result is in line previous studies for European and OECD countries (OECD, 2012).

Thus, the percentage of firms that have introduced innovations in the crisis period

has been reduced drastically, especially in the micro and small enterprises. In times

of crisis, under the mix of uncertainty and risk, it makes difficult for small firms to

access external capital markets to finance innovation projects (B. H. Hall, Moncada-

Paternò-Castello, Montresor, & Vezzani, 2016; Lee, Sameen, & Cowling, 2015). Yet,

for firms that decide to innovate, the crisis does not diminish the intensity in

innovation spending (especially medium and large companies). Therefore, size

becomes a key variable for understanding the decision and how much to spend on

innovation.

22

Previous studies show a mix of results on the effects of size in the decision to

innovate and how much to invest (e.g. Lööf & Heshmati, 2006). In this research, we

found that the size of the firm reduces the probability of investing in innovation in pre-

crisis periods, but not in contractionary periods. This implies that small firms have

greater difficulty in embarking on innovation activities in the face of uncertain

scenarios such as periods of crisis. On the other hand, the larger the size of the

company, the lower the intensity of innovation costs, regardless of the

macroeconomic conditions. However, in our results, the size is not related to the

outputs of the innovation. One possible explanation is that the measure of innovation

output used (the share of sales of new products or services for the company or

market) underestimates the share when total sales are high, as in large companies.

Another is that in medium and larger firm transaction and bureaucratic costs can

reduce the efficiency of innovations.

Experience and persistence in innovation to generate "innovation output" in periods

of crisis.

One of the main results of the study is that the economic crisis has not only

negatively affected the probability of innovation, but also (1) innovation performance

and (2) the effect of these outputs have on company growth.

Although the relationship between innovation input (innovation investment) and

innovation output remains positive in both pre-crisis and crisis periods, the

relationship returns are slightly lower in periods of crisis. In this period, investing

more in innovation has associated a smaller increase of the performance of this

innovation than in pre-crisis periods. One possible explanation is the reduction of

expenditures on equipment and machinery for innovation, a variable that some

studies have found to be "crucial" to generate innovation output (Love & Roper, 2015;

Pellegrino, Piva, & Vivarelli, 2012). Another explanation is that liquidity and financing

constraints in times of crisis force the firm to keep only innovation projects that

preserve their relative position in the market compared to the competitors in terms of

technology and knowledge (Auh and Menguc, 2005; March, 1991).

However, in our study, two factors are important to generate innovation output in

contractive periods: previous experience in innovation and persistence in R&D. The

positive effect of experience is higher in times of crisis than in expansion. Experience

23

in related projects can create internal capabilities within the organization, learning

economies, and "internal spillovers" that reduce the negative side effects of the

economic crisis on innovation performance (Phene & Almeida, 2008; Teece, 2014).

This result is in line with Amore (2015) who obtained that companies with experience

in innovation during previous economic crises obtain greater yield of the innovation in

new crises.

With respect to persistence in the knowledge created, continuous R&D is creative

work carried out within the enterprise, which is undertaken systematically in order to

increase the volume of knowledge to conceive new applications, such as goods or

services and new or significantly improved processes. In our study, we find that

companies that carry out R&D on a continuous basis obtain greater performance

from their innovations than those companies that buy R&D or simply do it on an

occasional basis. This positive effect is most evident in times of crisis. These results

imply that those companies with a continuous innovation strategy (which is also

related to the size of the company), are capable of generating higher levels of

introduction of new products in periods of economic crisis. Under the evolutionary

perspective, firms that carry out R&D activities on a continuous basis accumulate

knowledge and extract technology and technological trajectories that help them to

improve innovation performance (Raymond et al., 2010). This continued learning is

especially relevant in times of crisis, where the company does not have the same

capacity to adapt to technological changes. The companies that innovate continually

have the capacity to adapt to the new circumstances of the environment and can

obtain greater yields of their innovation. Yet, the purchase of R&D is not enough to

generate innovation output in times of crisis. Internal and external R&D is one of the

most studied variables in the field of innovation with a positive relationship in

innovation output (e.g. Crepon et al, 1998; Love et al. 2009; Roper et al. 2008).

Therefore, companies that do not have previous experience on innovations or are

involved in innovation in an occasional basis must re-evaluate the performance of

their innovations in times of crisis so as not to fall into inefficiencies and sunk costs.

Firm growth: The role of the innovation outputs.

Apart from the negative effect of the period of crisis on firm growth, the results show

that the economic crisis limits the return of innovation. Innovation output positively

24

influences the growth of the company, but in periods of crisis the positive effect of

innovation output is reduced by half. This implies that a greater share of sales due to

the introduction of new products or services for the company or market does not

entail a clear increase in the firm growth. One possible explanation is that in crisis

there are adjustments in demand. Consumers have greater budget constraints, which

may limit the demand for innovative products. Another is that the innovations

introduced are not radical enough to generate a competitive advantage in times of

crisis. These recessive periods force firms to keep only the innovation projects that

allow them to maintain their relative position in the market compared to the

competitors, in terms of technology and knowledge (Auh & Menguc, 2005; March,

1991).

Is innovation recommendable in times of crisis? The answer is yes, innovation helps

the growth of the company through sales, but the firm must be aware that the

performance of innovation is lower than in expansive periods. Therefore, companies

should: First, recalculate the opportunity cost of innovation, second invest in projects

with previous experience, or to cooperate with companies experienced in innovation,

and third to focus efforts on continued R&D activities. All this together helps

companies in contractive periods to reduce risk and to take positive externalities of

the characteristics of the company on innovation.

The role of public funding in innovation.

Another main question is whether public funds earmarked for firm innovation are

effective in times of crisis. According to the results, public aid increases the

investment in innovation of firms significantly, regardless of the period considered. In

addition, public funds increase innovation output in times of crisis, but not in pre-crisis

periods. Specifically, public funds from Spanish administrations have a positive effect

on the introduction of new products and services in times of economic crisis. This

relationship is not seen in expansive periods or funds coming from the EU. One

explanation is that many of the EU funds are based on very long-term innovation

projects, with a fundamental objective of expanding the boundaries of knowledge and

generating positive externalities for society, thus its positive effect cannot be

captured with this model. In any case, public subsidies allow firms to increase their

spending on innovation and generate greater outputs of this investment in times of

25

crisis, especially in small and micro enterprises, which ends up influencing the growth

of the company. However, cooperation with public entities (e.g. Universities, or state

research centers), while increasing investment in innovation, decreases the share of

new products and services in the enterprise market (output) in times of crisis. Again,

one possible explanation is that innovations developed in cooperation with public

entities may be directed towards other types of innovation, such as process

innovations rather than products.

Despite the interest of these results, it is important to mention that this study was

subject to several limitations. First, the indicators used to measure firm growth at the

firm level are not neutral with respect to empirical results (Audretsch, Segarra, et al.,

2014). For comparability, we selected the variable most frequently used in this field,

but results may be different with other performance variables. Second, although we

corrected in the model for selection bias and omitted variables, firm growth, as well

as innovation output, it is also associated with specific unobservable firm’s

capabilities, such as managerial capacity, entrepreneurship, ownership or firm

diversification. In that sense, we cannot use other firm’s accounting measures

relevant to calculate firm growth due to firm’s anonymization of the sample. Finally,

the study is conducted in Spain, which may not represent other countries or specific

sectors.

26

APPENDIX A

Table A.1. Explanation of variables.

Dependent variables Abbreviation Definition Eq. (1): Decision to innovate Dummy variable; 1 if firm, in years t to t-2, engaged in any type of

innovation expenditures follow Oslo Manual 2005 definition. Thus is in intramural or extramural R&D, purchased new machinery, equipment, software or other external knowledge, engaged in training of personnel, market research or did any other preparations to implement new or significantly improved products and processes, or new or significantly improved organization or distribution methods.

Eq. (2): Innovation input (natural logarithm)

Inno. input Innovation intensity: Amount (in euro) of average expenditure on innovation in year t to t-2 divided by the average employees in the period t to t-2 (Innovation expenditure: Oslo Manual 2005 definition).

Eq. (3): Innovation output (natural logarithm)

Inno. output Percent of firm’s turnover in year t coming from goods or services that were new to market or to the firm in years t to t-2.

Eq. (4): Firm growth (natural logarithm)

Sales growth Sales growth: Index number with base in firm's turnover in 2005

Independent variables Firm size (natural logarithm) Firm size Number of employees in year t Private company with foreign participation

PRIV Dummy variable; 1 if firm is private and with foreign participation in year t.

Company age (natural logarithm)

Age Number of years since the set-up of the firm

Gross investment (natural logarithm)

GITA Gross investment (in euros) in tangible assets in year t

Part of a group Part. group Dummy variable; 1 if firm is part of an enterprise in year t. Part of a group being the parent

Parent Dummy variable; 1 if firm is part of an enterprise group being the parent company in year t.

Part of a group being the subsidiary

Subsidiary Dummy variable; 1 if firm is part of an enterprise group being a subsidiary company in year t.

High degree employees High degree Percent of employees in the firm with University studies. Location in a Scientific or Technological Park

Tech. park Dummy variable; 1 if firm is located in a Scientific or Technological Park in year t

Exporter abroad UE Export. Categorical variable: 1 if firm export outside the UE less than 20% of the turnover; 2 if firm export in range 20% to 50%; 3 if firm export more than 50%; 0 in other cases (in year t).

Exporter in t-3 Exp. t-2 Dummy variable; 1 if firm sold goods or services outside the UE in year t-2

Innovation characteristics Intramural R&D Intra. R&D Dummy variable; 1 if firm is involved in-house R&D in year’s t to t-2. Continuous R&D Cont. R&D Dummy variable; 1 if firm is involved in continuous in-house R&D in

year t. (lagged in equation 3) Survey question: “Has your company carried out internal R&D activities in year t? Continuously or occasionally?”

Extramural R&D Entra. R&D Dummy variable; 1 if firm bought R&D from other enterprise or research organization in years t. (lagged in equation 3)

Process innovation Proc. inno Dummy variable; 1 if firm introduced process innovation in year’s t to t-2

Non-technological innovation N-T. inn Dummy variable; 1 if firm in years t to t-2, introduced organizational or marketing innovations follow Oslo Manual 2005 definition.

Explorative innovation Expl. Inn. Dummy variable; 1 if firm considered as highly important factor in their decision to innovate (t; t-2) at least two of the following: (i) increase range of goods or services; (ii) entering new markets; (iii) increased market share

27

Accumulative innovation Accum. Inn. Dummy variable; 1 if firm considered as highly important factor in their decision to innovate (t; t-2) at least half of the following: (i) Replace old products and process; (ii) Improving quality of goods or services; (iii) Improving flexibility for producing goods or services; (iv) Increasing capacity for producing goods or services; (v) Reducing costs per unit produced; (vi) Less energy per unit produced; (vii) Less materials per unit produced

Expenses in equipment and software (natural logarithm)

Soft. Percentage of total innovation expenditure in t, on acquisition of machines, equipment and software (not included in R&D).

Experience in innovation Inno. in t-3 Dummy variable; 1 if firm was able to introduce an innovation in t-3. Hampering innovation Cost factors Cost f. Dummy variable; 1 if firm perceives the lack of funds, finance from

sources outside the enterprise and high costs of innovation as highly important factors hampering its innovation activities, projects or decision to innovate (years t to t-2).

Knowledge factors Know. F. Dummy variable; 1 if firm perceives the lack of qualified personnel, information on technology or markets or difficulties in finding cooperation partners for innovation as highly important factors hampering its innovation activities, projects or decision to innovate (years t to t-2).

Market factors Market f. Dummy variable; 1 if firm perceives the domination over market by established enterprises or the uncertainty of demand for innovation goods and services as highly important factors hampering its innovation activities, projects or decision to innovate (years t to t-2).

Public support Public financial support Public fund Dummy variable; 1 if firm received financial support for innovation

activities from local/regional, central government or EU authorities (loans, grants, subsidies ...) in year t to t-2

National subsidies Spa. funds Dummy variable; 1 if firm in years t to t-2 received financial support for innovation activities from central government. (with 1 lag in Equation 3)

EU subsidies UE funds Dummy variable; 1 if firm in years t to t-2 received financial support for innovation activities from EU authorities (with 1 lag in Equation 3)

Sources of innovation about technological innovations Internal sources Internal S. Dummy variable; 1 if firm perceives sources of information within

enterprise or group as highly important sources of information on innovation (t; t-2).

Market sources Market S. Dummy variable; 1 if firm perceives suppliers, customers, competitors, consultants or R&D labs as highly important sources of information on innovation (t; t-2).

Institutional sources Inst. S. Dummy variable; 1 if firm perceives universities or government as highly important sources of information on innovation (t; t-2).

Cooperation Cooperation with abroad countries

Coop. abr. Dummy variable; 1 if firm cooperated on innovations with other overseas enterprises or institutions in years t to t-2

Cooperation with competitors Competitors Dummy variable; 1 if firm cooperated on innovations with other competitors enterprises or enterprises in the same main sector in years t to t-2 prior to survey

Cooperation with public institutions

Coop. Public Dummy variable; 1 if firm cooperated on innovations with public institutions in years t to t-2 prior to survey

Firm main sector Food, beverages and tobacco Food Dummy variable; 1 if firm operates in manufacture of food, beverages

and tobacco products. Textile Textile Dummy variable; 1 if firm operates in manufacture of textiles Chemical Chemical Dummy variable; 1 if firm operates in manufacture of chemicals and

chemical products.

28

Textile Dummy variable; 1 if firm operates in manufacture of textiles Water and energy W&E Dummy variable; 1 if firm operates in electricity, gas, steam, air

conditioning supply, water supply; sewerage; waste management and remediation activities (Group D and E NACE2009)

High Tech Manufactures HT manu. Dummy variable; 1 if firm operates in High Technology manufacture sector (OCDE Definition)

High Tech Services HT serv. Dummy variable; 1 if firm operates in High Technology service sector (OCDE Definition)

Construction Const. Dummy variable; 1 if firm operates in construction. (Group F) Hospitality Hospitality Dummy variable; 1 if firm operates in accommodation and food

service activities (Group I) Finance Finance Dummy variable; 1 if firm operates in financial and insurance sector

(Group K) Health Health Dummy variable; 1 if firm operates in human health and social work

activities (Group Q) Controls IMR equation 1 IM.1 Inverse Mill’s ratio from selection equation IMR auxiliary equation 3 IM.3 Inverse Mill’s ratio from auxiliary equation of innovation output with

dependent variable Dummy variable be 1 if firm in year t to t-2, introduced process, organizational or marketing innovations but not product innovation.

Period of crisis Dummy variable; 1 in the period 2009 to 2013

APPENDIX B

Figure B.1. Histogram of the ratio of innovation expenditure per employee (in

logarithms). 0

.2.4

.6.8

Den

sity

0 5 10 15Innovation expenditure per employee (log)

APPENDIX C

Table C1. Results of the selection equation.

Total firms sample Micro and small firms sample Medium and large firms sample Total Pre-crisis Crisis Total Pre-crisis Crisis Total Pre-crisis Crisis Factors hampering innovation Cost f.a 0.297*** 0.362*** 0.267*** 0.227*** 0.326*** 0.179*** 0.403*** 0.450*** 0.387*** (0.0208) (0.0313) (0.0265) (0.028) (0.043) (0.033) (0.032) (0.047) (0.038) Know. Fa -0.0183 -0.0108 -0.0419 -0.048 -0.063 -0.044 0.010 0.055 -0.031 (0.0276) (0.0395) (0.0359) (0.035) (0.051) (0.041) (0.043) (0.061) (0.054) Market f.a 0.0475** 0.0218 0.0651** 0.007 0.013 0.003 0.119*** 0.075 0.145*** (0.0230) (0.0338) (0.0295) (0.030) (0.045) (0.035) (0.036) (0.051) (0.044) Previous experiences in innovation Inno. in t-3a 1.162*** 1.253*** 1.105*** 0.988*** 1.013*** 0.976*** 1.318*** 1.520*** 1.155*** (0.0229) (0.0312) (0.0308) (0.033) (0.046) (0.038) (0.032) (0.043) (0.042) Firms characteristics Firm sizeb -0.0690*** -0.184*** 0.0150 (0.00935) (0.0127) (0.0119) Company ageb -0.0855*** -0.0636** -0.0763** -0.212*** -0.232*** -0.195*** -0.001 -0.013 0.012 (0.0222) (0.0267) (0.0302) (0.031) (0.038) (0.039) (0.030) (0.035) (0.038) GITAb 0.0569*** 0.0533*** 0.0592*** 0.065*** 0.059*** 0.068*** 0.044*** 0.038*** 0.049*** (0.00171) (0.00239) (0.00226) (0.002) (0.004) (0.003) (0.002) (0.003) (0.003) Type of company PRIV a 0.00672 0.0144 -0.0291 -0.077 -0.316*** -0.004 -0.001 0.049 -0.027 (0.0385) (0.0549) (0.0464) (0.075) (0.119) (0.081) (0.045) (0.062) (0.054) Parenta 0.251*** 0.314*** 0.217*** 0.148* 0.112 0.158* 0.247*** 0.251*** 0.261*** (0.0466) (0.0679) (0.0570) (0.082) (0.128) (0.087) (0.055) (0.077) (0.067) Subsidiarya 0.0463 0.0574 0.0308 0.011 0.192** -0.053 0.038 -0.057 0.108** (0.0314) (0.0430) (0.0380) (0.053) (0.085) (0.059) (0.038) (0.051) (0.046) Firm environment Tech. parka 0.605*** 0.617*** 0.635*** 0.525*** 0.593*** 0.508*** 0.714*** 0.619*** 0.753*** (0.0742) (0.117) (0.0893) (0.093) (0.145) (0.105) (0.123) (0.187) (0.149) Export in t-2a 0.280*** 0.314*** 0.263*** 0.233*** 0.276*** 0.218*** 0.335*** 0.341*** 0.336*** (0.0176) (0.0272) (0.0208) (0.023) (0.041) (0.025) (0.026) (0.036) (0.031)

31

(continued) Total firms sample Micro and small firms sample Medium and high firms sample Total Pre-crisis Crisis Total Pre-crisis Crisis Total Pre-crisis Crisis Crisisa -0.520*** -0.676*** -0.337*** (0.0182) (0.025) (0.027) Firm main sector Constructiona -0.232*** -0.171** -0.247*** -0.470*** -0.425*** -0.481*** -0.117* -0.108 -0.126 (0.0585) (0.0791) (0.0740) (0.095) (0.128) (0.105) (0.070) (0.098) (0.089) Healtha -0.0565 0.00958 -0.167* 0.134 0.224* 0.035 -0.156** -0.135 -0.169* (0.0602) (0.0692) (0.0907) (0.108) (0.121) (0.165) (0.072) (0.089) (0.096) W&Ea 0.153* 0.198 0.112 0.025 0.025 0.027 0.219* 0.319* 0.182 (0.0895) (0.148) (0.114) (0.145) (0.229) (0.153) (0.117) (0.192) (0.138) Hospitalitya -1.072*** -1.203*** -0.928*** -1.104*** -0.900*** -1.303*** -1.014*** -1.274*** -0.831*** (0.0997) (0.140) (0.133) (0.215) (0.244) (0.370) (0.113) (0.177) (0.133) Fooda -0.0166 0.125* -0.120** -0.229*** -0.065 -0.316*** 0.196*** 0.328*** 0.124 (0.0444) (0.0644) (0.0573) (0.062) (0.091) (0.074) (0.065) (0.090) (0.084) Textilea 0.119 0.249* 0.0405 -0.070 0.007 -0.112 0.466*** 0.672*** 0.343** (0.0874) (0.138) (0.103) (0.109) (0.165) (0.122) (0.162) (0.261) (0.168) HT manu.a 0.612*** 0.511*** 0.703*** 0.510*** 0.441*** 0.557*** 0.903*** 0.727*** 1.156*** (0.0757) (0.110) (0.0953) (0.088) (0.127) (0.100) (0.152) (0.205) (0.269) HT serv.a 0.448*** 0.675*** 0.393*** 0.567*** 1.071*** 0.489*** 0.311*** 0.508** 0.244** (0.0723) (0.146) (0.0805) (0.094) (0.207) (0.098) (0.109) (0.204) (0.119) Constantb -0.0161 0.350*** -0.883*** 0.440*** 0.429*** -0.247* -0.830*** -0.803*** -1.162*** (0.0687) (0.0861) (0.0968) (0.098) (0.119) (0.128) (0.100) (0.121) (0.134) Sample size 59,949 26,644 26,644 30,436 13,289 17,147 29,513 13,355 16,158 Number of firms 6661 6661 6661 3758 3489 3648 3658 3518 3464 Note: a: Dummy variable; b: Continuous variable in logarithmic; c: Ordered categorical variable. Cluster-Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. See Appendix A for variable details.

32

Table C2. Results of the innovation investment equation. Total firms sample Micro and small firms sample Medium and large firms sample Total Pre-

crisis Crisis Crisis a. Total Pre-

crisis Crisis Crisis a. Total Pre-crisis Crisis Crisis a.

Highly important sources of information about innovation Internal s.a 0.208*** 0.215*** 0.193*** 0.169*** 0.166*** 0.180*** 0.147*** 0.137*** 0.225*** 0.234*** 0.206*** 0.182*** (0.020) (0.024) (0.025) (0.025) (0.025) (0.031) (0.032) (0.032) (0.034) (0.042) (0.042) (0.043) Market s.a 0.143*** 0.145*** 0.140*** 0.114*** 0.116*** 0.089*** 0.137*** 0.124*** 0.165*** 0.182*** 0.149*** 0.127*** (0.019) (0.024) (0.024) (0.024) (0.024) (0.030) (0.030) (0.030) (0.033) (0.041) (0.040) (0.040) Inst. S.a 0.078*** 0.063** 0.091*** 0.087*** 0.088** 0.079* 0.099** 0.097** 0.121*** 0.112** 0.120** 0.117** (0.027) (0.032) (0.032) (0.033) (0.035) (0.043) (0.043) (0.043) (0.044) (0.055) (0.051) (0.051) Factors hampering innovation Cost factorsa -0.033* -0.004 -0.056** -0.062** 0.017 0.052 -0.014 -0.017 -0.073** -0.033 -0.099** -0.101** (0.020) (0.025) (0.026) (0.026) (0.026) (0.032) (0.034) (0.034) (0.033) (0.042) (0.041) (0.041) Know. F. a

-0.074*** -

0.084*** -0.066** -0.072** -0.076** -0.095** -0.060 -0.065* -0.039 -0.033 -0.051 -0.054 (0.024) (0.028) (0.031) (0.031) (0.030) (0.037) (0.039) (0.039) (0.044) (0.052) (0.058) (0.058) Market

factorsa -0.079*** -

0.091*** -0.071*** -

0.077*** -

0.078*** -0.063* -

0.091*** -

0.092*** 0.007 -0.031 0.034 0.027 (0.020) (0.025) (0.025) (0.025) (0.026) (0.033) (0.032) (0.032) (0.035) (0.045) (0.044) (0.044) Type of innovation Intramural

R&Da 0.923*** 0.842*** 0.998*** 0.983*** 0.837*** 0.770*** 0.908*** 0.903*** 0.989*** 0.893*** 1.068*** 1.053*** (0.024) (0.027) (0.030) (0.030) (0.029) (0.035) (0.037) (0.037) (0.041) (0.048) (0.053) (0.053) Extramural

R&Da 0.472*** 0.471*** 0.471*** 0.469*** 0.356*** 0.344*** 0.368*** 0.368*** 0.511*** 0.515*** 0.502*** 0.500*** (0.022) (0.027) (0.028) (0.028) (0.028) (0.033) (0.036) (0.036) (0.036) (0.044) (0.043) (0.043) Soft.c 0.195*** 0.487*** -0.049 -0.057 0.132** 0.457*** -0.131* -0.139* 0.177** 0.470*** -0.085 -0.083 (0.047) (0.060) (0.058) (0.058) (0.060) (0.082) (0.074) (0.074) (0.075) (0.091) (0.093) (0.093) Expl. Inn.a 0.097*** 0.012 0.176*** (0.028) (0.036) (0.045) Accum. Inn.a 0.121*** 0.091** 0.028 (0.029) (0.038) (0.043) Cooperation with: Coop. Abr.a 0.299*** 0.334*** 0.311*** 0.308*** 0.172*** 0.208*** 0.163*** 0.162*** 0.195*** 0.169*** 0.257*** 0.255*** (0.032) (0.043) (0.036) (0.036) (0.044) (0.055) (0.053) (0.053) (0.044) (0.061) (0.048) (0.048)

33

Coop. Public a 0.111*** 0.053* 0.149*** 0.152*** 0.136*** 0.094** 0.163*** 0.164*** 0.048 0.028 0.057 0.059 (0.025) (0.031) (0.030) (0.030) (0.033) (0.040) (0.042) (0.042) (0.038) (0.050) (0.045) (0.045) Competitorsa 0.277*** 0.176*** 0.332*** 0.330*** 0.162*** 0.138** 0.172*** 0.172*** 0.235*** 0.124* 0.297*** 0.292*** (0.035) (0.045) (0.040) (0.040) (0.042) (0.058) (0.049) (0.049) (0.053) (0.070) (0.060) (0.060) Access to subsidies Public fundsa 0.633*** 0.626*** 0.616*** 0.617*** 0.642*** 0.623*** 0.630*** 0.633*** 0.583*** 0.572*** 0.578*** 0.577*** (0.021) (0.025) (0.027) (0.027) (0.028) (0.033) (0.037) (0.037) (0.033) (0.043) (0.041) (0.041) Type of company PRIVa 0.200*** 0.188*** 0.201*** 0.202*** 0.083 -0.022 0.131** 0.137** 0.055 0.069 0.046 0.043 (0.042) (0.053) (0.046) (0.046) (0.061) (0.085) (0.065) (0.066) (0.052) (0.063) (0.059) (0.059) Parenta 0.257*** 0.300*** 0.215*** 0.216*** 0.083 0.109 0.062 0.064 -0.104* -0.089 -0.120* -0.122* (0.044) (0.055) (0.050) (0.050) (0.072) (0.092) (0.079) (0.079) (0.061) (0.074) (0.069) (0.069) Subsidiarya 0.288*** 0.353*** 0.231*** 0.229*** -0.020 0.073 -0.085 -0.088 0.122*** 0.133** 0.109** 0.106** (0.032) (0.041) (0.037) (0.037) (0.050) (0.064) (0.057) (0.057) (0.045) (0.055) (0.052) (0.052) (continued) Total firms simple Micro and small firms sample Medium and large firms sample Total Pre-

crisis Crisis Crisis a. Total Pre-

crisis Crisis Crisis a. Total Pre-crisis Crisis Crisis a.

Firms characteristics and environment Firm sizeb

-0.581*** -

0.607*** -0.558*** -

0.560***

(0.013) (0.014) (0.015) (0.015) Export c 0.053*** 0.060*** 0.045*** 0.042*** -0.014 -0.016 -0.019 -0.019 0.124*** 0.101*** 0.143*** 0.137*** (0.013) (0.015) (0.016) (0.016) (0.018) (0.022) (0.021) (0.021) (0.021) (0.024) (0.026) (0.026) Company

ageb -0.027 -0.026 -0.024 -0.024 -

0.398*** -

0.372*** -

0.453*** -

0.452*** -0.009 -0.010 -0.002 -0.000 (0.024) (0.025) (0.031) (0.031) (0.030) (0.031) (0.039) (0.039) (0.035) (0.037) (0.043) (0.043) GITAb

0.018*** 0.020*** 0.015*** 0.015*** 0.005* 0.009** 0.001 0.001 -

0.015*** -0.014*** -

0.017*** -0.017*** (0.002) (0.003) (0.003) (0.003) (0.003) (0.004) (0.003) (0.003) (0.004) (0.005) (0.005) (0.005) Tech. parka 0.373*** 0.420*** 0.334*** 0.336*** 0.330*** 0.374*** 0.295*** 0.297*** 0.421*** 0.585*** 0.311*** 0.306*** (0.052) (0.069) (0.056) (0.056) (0.064) (0.086) (0.069) (0.069) (0.095) (0.128) (0.098) (0.098) Crisisa 0.095***

0.150*** 0.154***

(0.017)

(0.023) (0.027) Firm main sector Const.a

-0.425*** -

0.389*** -0.489*** -

0.479*** -

0.394*** -

0.333*** -

0.481*** -

0.478*** -

0.612*** -0.630*** -

0.634*** -0.621*** (0.078) (0.094) (0.095) (0.094) (0.104) (0.128) (0.141) (0.141) (0.126) (0.150) (0.142) (0.141) Healtha -0.263*** -0.206** -0.371*** - 0.316*** 0.427*** -0.046 -0.042 - -0.952*** - -0.720***

34

0.357*** 0.848*** 0.732*** (0.083) (0.091) (0.116) (0.116) (0.109) (0.119) (0.180) (0.180) (0.132) (0.148) (0.162) (0.162) W&Ea 0.052 0.259** -0.078 -0.069 0.179 0.360* 0.096 0.092 -0.201 -0.018 -0.307* -0.283* (0.108) (0.128) (0.127) (0.127) (0.194) (0.204) (0.253) (0.252) (0.146) (0.176) (0.170) (0.170) Hospitalitya

-1.094*** -

1.154*** -1.041*** -

1.046*** -

1.645*** -

1.822*** -

1.337*** -

1.336*** -

0.911*** -0.900** -

0.893*** -0.898*** (0.214) (0.327) (0.237) (0.234) (0.227) (0.292) (0.167) (0.164) (0.261) (0.360) (0.299) (0.296) Fooda

-0.178*** -

0.185*** -0.190*** -

0.190*** -

0.233*** -0.180** -

0.310*** -

0.310*** -0.102 -0.145* -0.076 -0.076 (0.048) (0.054) (0.058) (0.058) (0.069) (0.085) (0.082) (0.081) (0.071) (0.079) (0.084) (0.084) Textilea -0.189** -0.159** -0.233** -0.233** -0.110 -0.070 -0.153 -0.151 -0.051 -0.001 -0.118 -0.117 (0.074) (0.076) (0.093) (0.092) (0.108) (0.115) (0.139) (0.139) (0.109) (0.114) (0.138) (0.137) HT. Manu.a 0.441*** 0.388*** 0.514*** 0.517*** 0.304*** 0.262*** 0.378*** 0.379*** 0.660*** 0.667*** 0.666*** 0.671*** (0.051) (0.055) (0.060) (0.060) (0.054) (0.063) (0.065) (0.065) (0.077) (0.083) (0.093) (0.093) H.T. Serv. a 0.910*** 1.034*** 0.844*** 0.862*** 0.853*** 0.962*** 0.775*** 0.780*** 1.219*** 1.377*** 1.158*** 1.181*** (0.059) (0.080) (0.061) (0.061) (0.065) (0.084) (0.069) (0.069) (0.129) (0.201) (0.125) (0.126) IM.1

-0.523*** -

0.488*** -0.521*** -

0.510*** -

0.276*** -0.194** -

0.305*** -

0.302*** -

1.127*** -1.202*** -

1.055*** -1.049*** (0.048) (0.071) (0.057) (0.056) (0.064) (0.098) (0.074) (0.074) (0.083) (0.123) (0.098) (0.098) Constant 9.005*** 9.031*** 9.068*** 9.070*** 8.562*** 8.387*** 8.965*** 8.962*** 6.447*** 6.433*** 6.598*** 6.588*** (0.073) (0.079) (0.101) (0.101) (0.097) (0.107) (0.139) (0.139) (0.134) (0.147) (0.180) (0.180) Sample size Number of firms

59,949 26,644 26,644 26,644 30,436 13,289 17,147 17,147 29,513 13,355 16,158 16,158 6661 6661 6661 6661 3758 3489 3648 3648 3658 3518 3464 3464

Note: a: Dummy variable; b: Continuous variable in logarithmic; c: Ordered categorical variable. IMR: Inverse Mills Ration. Cluster-Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. See Appendix A for variable details.

REFERENCES

Aghion, P., Askenazy, P., Berman, N., Cette, G., & Eymard, L. (2012). Credit

Constraints and the cyclicality of R&D investment: Evidence from France.

Journal of the European Economic Association, 10(5), 1001-1024.

doi:10.1111/j.1542-4774.2012.01093.x

Amore, M. D. (2015). Companies learning to innovate in recessions. Research

Policy, 44(8), 1574-1583. doi:https://doi.org/10.1016/j.respol.2015.05.006

Antonioli, D., Bianchi, A., Mazzanti, M., Montresor, S., & Pini, P. (2013). Innovation

strategies and economic crisis: Evidence from firm-level Italian data.

Economia politica, 30(1), 33-68.

Archibugi, D., Filippetti, A., & Frenz, M. (2013a). Economic crisis and innovation: Is

destruction prevailing over accumulation? Research Policy, 42(2), 303-314.

doi:https://doi.org/10.1016/j.respol.2012.07.002

Archibugi, D., Filippetti, A., & Frenz, M. (2013b). The impact of the economic crisis on

innovation: Evidence from Europe. Technological Forecasting and Social

Change, 80(7), 1247-1260. doi:https://doi.org/10.1016/j.techfore.2013.05.005

Arrow, K. J. (1962). The Economic Implications of Learning by Doing. The Review of

Economic Studies, 29(3), 155-173. doi:10.2307/2295952

Audretsch, D. B., Coad, A., & Segarra, A. (2014). Firm growth and innovation. Small

Business Economics, 43(4), 743-749. doi:10.1007/s11187-014-9560-x

Audretsch, D. B., & Feldman, M. P. (2004). Chapter 61 - Knowledge Spillovers and

the Geography of Innovation. In J. V. Henderson & T. Jacques-François

(Eds.), Handbook of Regional and Urban Economics (Vol. Volume 4, pp.

2713-2739): Elsevier.

Audretsch, D. B., Segarra, A., & Teruel, M. (2014). Why don't all young firms invest in

R&D? Small Business Economics, 43(4), 751-766. doi:10.1007/s11187-014-

9561-9

Auh, S., & Menguc, B. (2005). Balancing exploration and exploitation: The

moderating role of competitive intensity. Journal of Business Research,

58(12), 1652-1661. doi:https://doi.org/10.1016/j.jbusres.2004.11.007

Barlevy, G. (2007). On the cyclicality of research and development. The American

Economic Review, 97(4), 1131-1164.

36

Bayus, B. L., Erickson, G., & Jacobson, R. (2003). The Financial Rewards of New

Product Introductions in the Personal Computer Industry. Management

Science, 49(2), 197-210. doi:10.1287/mnsc.49.2.197.12741

Bessler, W., & Bittelmeyer, C. (2008). Patents and the performance of technology

firms: Evidence from initial public offerings in Germany. Financial Markets and

Portfolio Management, 22(4), 323-356. doi:10.1007/s11408-008-0089-3

Bovha Padilla, S., Damijan, J. P., & Konings, J. (2009). Financial Constraints and the

Cyclicality of R&D investment: Evidence from Slovenia.

Breschi, S., Malerba, F., & Orsenigo, L. (2000). Technological regimes and

Schumpeterian patterns of innovation. The Economic Journal, 110(463), 388-

410.

Burke, W. J. (2009). Fitting and interpreting Cragg's tobit alternative using Stata.

Stata Journal, 9(4), 584.

Choi, S. B., & Williams, C. (2014). The impact of innovation intensity, scope, and

spillovers on sales growth in Chinese firms. Asia Pacific Journal of

Management, 31(1), 25-46. doi:10.1007/s10490-012-9329-1

Coad, A., & Rao, R. (2010). Firm growth and R&D expenditure. Economics of

Innovation and New Technology, 19(2), 127-145.

doi:10.1080/10438590802472531

Coad, A., Segarra, A., & Teruel, M. (2016). Innovation and firm growth: Does firm

age play a role? Research Policy, 45(2), 387-400.

doi:http://dx.doi.org/10.1016/j.respol.2015.10.015

Cohen, W. M., & Klepper, S. (1996). A Reprise of Size and R&D. The Economic

Journal, 106(437), 925-951. doi:10.2307/2235365

Cragg, J. G. (1971). Some Statistical Models for Limited Dependent Variables with

Application to the Demand for Durable Goods. Econometrica, 39(5), 829-844.

doi:10.2307/1909582

Crepon, B., Duguet, E., & Mairessec, J. (1998). Research, Innovation And

Productivity: An Econometric Analysis at The Firm Level. Economics of

Innovation and New Technology, 7(2), 115-158.

doi:10.1080/10438599800000031

Dachs, B., & Peters, B. (2014). Innovation, employment growth, and foreign

ownership of firms: A European perspective. Research Policy, 43(1), 214-232.

doi:https://doi.org/10.1016/j.respol.2013.08.001

37

Eakins, J. (2016). An application of the double hurdle model to petrol and diesel

household expenditures in Ireland. Transport Policy, 47, 84-93.

doi:https://doi.org/10.1016/j.tranpol.2016.01.005

Engel, C., & Moffatt, P. G. (2014). dhreg, xtdhreg, and bootdhreg: Commands to

implement double-hurdle regression. Stata Journal, 14(4), 778-797.

Filippetti, A., & Archibugi, D. (2011). Innovation in times of crisis: National Systems of

Innovation, structure, and demand. Research Policy, 40(2), 179-192.

doi:https://doi.org/10.1016/j.respol.2010.09.001

Freeman, C. (1994). The economics of technical change. Cambridge Journal of

Economics, 18(5), 463-514.

Griffith, R., Huergo, E., Mairesse, J., & Peters, B. (2006). Innovation and Productivity

Across Four European Countries. Oxford Review of Economic Policy, 22(4),

483-498. doi:10.1093/oxrep/grj028

Griliches, Z. (1986). Productivity, R and D, and Basic Research at the Firm Level in

the 1970's. The American Economic Review, 76(1), 141-154.

Hall, B. H., Mairesse, J., & Mohnen, P. (2010). Chapter 24 - Measuring the Returns

to R&D. In H. H. Bronwyn & R. Nathan (Eds.), Handbook of the Economics of

Innovation (Vol. Volume 2, pp. 1033-1082): North-Holland.

Hall, B. H., Moncada-Paternò-Castello, P., Montresor, S., & Vezzani, A. (2016).

Financing constraints, R&D investments and innovative performances: new

empirical evidence at the firm level for Europe. Economics of Innovation and

New Technology, 25(3), 183-196. doi:10.1080/10438599.2015.1076194

Hall, B. H., & Sena, V. (2017). Appropriability mechanisms, innovation, and

productivity: evidence from the UK. Economics of Innovation and New

Technology, 26(1-2), 42-62. doi:10.1080/10438599.2016.1202513

Hall, R. E. (1991). Recessions as reorganizations. NBER macroeconomics annual,

17-47.

Harrison, R., Jaumandreu, J., Mairesse, J., & Peters, B. (2014). Does innovation

stimulate employment? A firm-level analysis using comparable micro-data

from four European countries. International Journal of Industrial Organization,

35, 29-43. doi:https://doi.org/10.1016/j.ijindorg.2014.06.001

Hashi, I., & Stojčić, N. (2013). The impact of innovation activities on firm performance

using a multi-stage model: Evidence from the Community Innovation Survey 4.

38

Research Policy, 42(2), 353-366.

doi:https://doi.org/10.1016/j.respol.2012.09.011

Hausman, A., & Johnston, W. J. (2014). The role of innovation in driving the

economy: Lessons from the global financial crisis. Journal of Business

Research, 67(1), 2720-2726.

doi:http://dx.doi.org/10.1016/j.jbusres.2013.03.021

Heckman, J. J. (1979). Sample Selection Bias as a Specification Error.

Econometrica, 47(1), 153-161. doi:10.2307/1912352

Klomp, L., & Van Leeuwen, G. (2001). Linking Innovation and Firm Performance: A

New Approach. International Journal of the Economics of Business, 8(3), 343-

364. doi:10.1080/13571510110079612

Lee, N., Sameen, H., & Cowling, M. (2015). Access to finance for innovative SMEs

since the financial crisis. Research Policy, 44(2), 370-380.

doi:https://doi.org/10.1016/j.respol.2014.09.008

Li, H., & Atuahene-Gima, K. (2001). Product Innovation Strategy and the

Performance of New Technology Ventures in China. Academy of Management

Journal, 44(6), 1123-1134. doi:10.2307/3069392

Lööf, H., & Heshmati, A. (2002). Knowledge capital and performance heterogeneity:

A firm-level innovation study. International Journal of Production Economics,

76(1), 61-85. doi:https://doi.org/10.1016/S0925-5273(01)00147-5

Lööf, H., & Heshmati, A. (2006). On the relationship between innovation and

performance: A sensitivity analysis. Economics of Innovation and New

Technology, 15(4-5), 317-344. doi:10.1080/10438590500512810

Love, J. H., & Roper, S. (2015). SME innovation, exporting and growth: A review of

existing evidence. International Small Business Journal, 33(1), 28-48.

doi:doi:10.1177/0266242614550190

Mairesse, J., & Robin, S. (2017). Assessing measurement errors in the CDM

research–innovation–productivity relationships. Economics of Innovation and

New Technology, 26(1-2), 93-107. doi:10.1080/10438599.2016.1210771

March, J. G. (1991). Exploration and exploitation in organizational learning.

Organization Science, 2(1), 71-87.

Miller, D. J. (2006). Technological diversity, related diversification, and firm

performance. Strategic Management Journal, 27(7), 601-619.

doi:10.1002/smj.533

39

Mohnen, P., & Hall, B. H. (2013). Innovation and Productivity: An Update. Eurasian

Business Review, 3(1), 47-65. doi:10.14208/bf03353817

OECD. (2005). The Measurement of Scientific and Technological Activities:

Guidelines for Collecting and Interpreting Innovation Data: Oslo Manual.

OECD. (2009). Policy Responses to the Economic Crisis: Investing in Innovation for

Long-Term Growth. Retrieved from

OECD. (2012). Innovation in the Crisis and Beyond. Retrieved from Paris:

OECD. (2015). Frascati Manual 2015: Guidelines for Collecting and Reporting Data

on Research and Experimental Development. Retrieved from

Ouyang, M. (2011). On the Cyclicality of R&D. The Review of Economics and

Statistics, 93(2), 542-553. doi:10.1162/REST_a_00076

Paunov, C. (2012). The global crisis and firms’ investments in innovation. Research

Policy, 41(1), 24-35. doi:https://doi.org/10.1016/j.respol.2011.07.007

Pellegrino, G., Piva, M., & Vivarelli, M. (2012). Young firms and innovation: A

microeconometric analysis. Structural Change and Economic Dynamics,

23(4), 329-340. doi:https://doi.org/10.1016/j.strueco.2011.10.003

Peters, B., Roberts, M. J., & Vuong, V. A. (2017). Dynamic R&D choice and the

impact of the firm's financial strength. Economics of Innovation and New

Technology, 26(1-2), 134-149. doi:10.1080/10438599.2016.1202516

Phene, A., & Almeida, P. (2008). Innovation in multinational subsidiaries: The role of

knowledge assimilation and subsidiary capabilities. Journal of International

Business Studies, 39(5), 901-919. doi:10.1057/palgrave.jibs.8400383

Rafferty, M., & Funk, M. (2008). Asymmetric effects of the business cycle on firm-

financed R&D. Economics of Innovation and New Technology, 17(5), 497-510.

doi:http://dx.doi.org/10.1080/10438590701407232

Raymond, W., Mohnen, P., Palm, F., & Loeff, S. S. v. d. (2010). Persistence of

Innovation in Dutch Manufacturing: Is It Spurious? The Review of Economics

and Statistics, 92(3), 495-504. doi:10.1162/REST_a_00004

Romer, P. M. (1990). Endogenous technological change. Journal of Political

Economy, 98(5, Part 2), S71-S102.

Segarra, A., & Teruel, M. (2014). High-growth firms and innovation: an empirical

analysis for Spanish firms. Small Business Economics, 43(4), 805-821.

doi:10.1007/s11187-014-9563-7

Stiglitz, J. E. (1993a). Endogenous growth and cycles. Retrieved from

40

Stiglitz, J. E. (1993b). The role of the state in financial markets. The World Bank

Economic Review, 7(suppl 1), 19-52.

Teece, D. J. (2014). A dynamic capabilities-based entrepreneurial theory of the

multinational enterprise. Journal of International Business Studies, 45(1), 8-37.

doi:10.1057/jibs.2013.54

Van Leeuwen, G., & Klomp, L. (2006). On the contribution of innovation to multi-

factor productivity growth. Economics of Innovation and New Technology,

15(4-5), 367-390.

Wooldridge, J. M. (2010). Econometric analysis of cross section and panel data (2nd

ed. ed.). Cambridge: MIT press.

TABLES

Table 1. Descriptive statistics on the decision to innovate during the period 2005 to 2013. Pre-crisis Crisis 2005 2006 2007 2008 2009 2010 2011 2012 2013 Percentage of innovative firms in: Product 55.16 54.30 52.15 53.63 55.58 56.31 44.12 39.38 37.71 Process 55.01 56.37 53.24 55.41 57.68 58.88 45.97 39.63 36.63 Organization and marketing

n.a n.a n.a 50.82 49.38 47.46 46.16 46.18 44.41

Total 72.51 72.51 70.45 76.49 77.74 78.56 68.34 65.35 63.31 Percentage of innovative firms by size: Micro firms 74.81 70.19 69.72 73.13 71.73 71.32 55.79 52.20 47.42 Small firms 79.49 79.04 75.19 80.65 83.03 82.99 71.34 67.16 65.06 Medium firms 76.75 78.06 75.21 81.18 82.81 83.99 75.84 73.83 73.39 Large firms 55.66 60.00 59.32 69.06 71.47 74.63 70.43 69.27 69.94

N.a: Not available;

Table 2. Descriptive statistics of innovation expenditure and output during the period 2005 to 2013.

Pre-crisis Crisis 2005 2006 2007 2008 2009 2010 2011 2012 2013 Average and median expenses in innovation (in logarithms) of the innovative firms: Micro firms 10.33 9.32 9.06 8.41 7.86 7.26 7.78 7.45 7.16 11.27 11.11 11.10 10.98 10.60 10.41 10.57 10.34 10.21 Small firms 10.83 10.26 10.06 9.59 9.13 8.71 9.00 8.96 8.93 11.70 11.67 11.72 11.55 11.43 11.38 11.45 11.42 11.50 Medium firms 11.55 11.06 10.84 10.36 10.39 10.03 10.52 10.10 10.00 12.37 12.34 12.44 12.39 12.37 12.32 12.49 12.43 12.44 Large firms 11.54 11.27 11.38 10.66 10.29 9.96 10.36 10.21 9.88 13.37 13.21 13.35 13.05 13.04 12.88 12.94 12.88 12.77 Percentage of expenditure on innovation for the acquisition of machines, equipment and software. Micro firms 18.59 19.40 17.70 15.26 15.92 14.00 14.52 9.36 10.11 Small firms 22.40 18.40 19.02 17.59 15.62 13.65 12.84 11.68 12.31 Medium firms 22.84 21.12 19.27 18.11 16.77 15.06 15.27 12.91 11.83 Large firms 28.54 24.96 24.37 23.21 27.15 22.49 21.48 19.71 18.06 Percentage of firm’s turnover in year t coming from goods or services that were new to market or to enterprise (only for companies that report product innovation): Micro firms 48.28 45.96 46.58 48.62 46.80 44.61 44.60 39.91 39.26 Small firms 41.60 41.57 40.87 44.47 41.62 41.85 41.72 39.64 40.50 Medium firms 38.75 38.27 36.43 37.74 37.16 36.64 39.52 38.84 39.31 Large firms 34.19 31.02 34.20 35.22 35.63 35.62 37.89 36.25 37.15

42

Table 3. Average and median firm growth (percentage) in different groups of firms in the database.

Pre-crisis Crisis 2005 2006 2007 2008 2009 2010 2011 2012 2013 Innovative firm vs non-innovative firm. (Average and median)

Non-Innovative

16.02 11.91 2.98 -8.79 3.16 3.32 -3.77 -1.88

6.89 6.87 0.01 -10.15 0.18 0.08 -4.71 -2.97

Innovative 17.03 16.14 7.27 -8.85 6.40 6.19 -2.48 0.64 8.77 8.58 1.59 -11.12 2.30 2.57 -3.71 -1.06

Exporter abroad UE vs non-exporter Non exporter

18.13 16.34 7.15 -7.11 2.86 3.69 -4.37 -2.36

8.14 8.18 2.05 -8.23 0.18 0.14 -4.73 -3.52 Exporter 14.75 12.67 4.77 -11.56 9.75 7.42 -1.14 2.23

8.18 7.72 -0.18 -14.66 5.01 3.76 -2.95 0.29

43

Table 4. Results of the innovation output equation.

Total firms sample Micro and small firms sample Medium and large firms sample Total Pre-crisis Crisis Crisis a. Total Pre-crisis Crisis Crisis a. Total Pre-crisis Crisis Crisis a. Innovation Characteristics Inno. inputb 0.244*** 0.256*** 0.234*** 0.230*** 0.220*** 0.260*** 0.196*** 0.202*** 0.133*** 0.193*** 0.114*** 0.122*** (0.034) (0.054) (0.043) (0.043) (0.031) (0.051) (0.040) (0.040) (0.029) (0.052) (0.035) (0.035) Cont. R&Da 0.228*** 0.189** 0.253*** 0.255*** 0.152** 0.140 0.168* 0.183** 0.228*** 0.305** 0.218** 0.218** (0.059) (0.091) (0.075) (0.075) (0.067) (0.106) (0.086) (0.086) (0.075) (0.130) (0.093) (0.092) Ext.. R&Da -0.043** -0.038 -0.045* -0.043* -0.034 -0.025 -0.041 -0.042 -0.022 -0.011 -0.017 -0.020 (0.018) (0.028) (0.023) (0.023) (0.026) (0.042) (0.034) (0.034) (0.028) (0.050) (0.034) (0.034) Proc. inno.a 0.066*** 0.079*** 0.060*** 0.037 0.097*** 0.084** 0.105*** 0.076** 0.037 0.103* 0.010 -0.022 (0.017) (0.028) (0.022) (0.024) (0.023) (0.038) (0.029) (0.032) (0.030) (0.053) (0.036) (0.038) N-T. inn.a 0.041* 0.071** -0.006 (0.024) (0.032) (0.036) Expl. inn.a 0.044* 0.021 0.065* (0.023) (0.033) (0.035) Accum. inn.a 0.109*** 0.100*** 0.135*** (0.0258) (0.037) (0.034) Inno. in t-3 a 0.264*** 0.189** 0.355*** 0.368*** 0.290*** 0.248** 0.357*** 0.385*** 0.126 0.124 0.184* 0.198* (0.060) (0.088) (0.082) (0.081) (0.070) (0.109) (0.095) (0.096) (0.078) (0.129) (0.103) (0.102) Cooperation Abroad inst.a 0.098*** 0.087 0.111** 0.117** 0.180*** 0.174** 0.175** 0.190*** 0.053 0.090 0.054 0.055 (0.037) (0.057) (0.047) (0.047) (0.056) (0.088) (0.073) (0.073) (0.050) (0.090) (0.061) (0.060) Public inst.a -0.060*** -0.025 -0.081*** -0.079*** -0.050 0.028 -0.091** -0.090** -0.092*** -0.088 -0.095** -0.090** (0.022) (0.034) (0.027) (0.027) (0.031) (0.050) (0.040) (0.040) (0.034) (0.061) (0.041) (0.040) Competitorsa -0.076*** -0.065 -0.077** -0.079** -0.128*** -0.113 -0.136** -0.142** -0.010 -0.021 -0.003 -0.008 (0.029) (0.047) (0.036) (0.035) (0.044) (0.074) (0.055) (0.055) (0.042) (0.078) (0.050) (0.049) Access to subsidies Spa. fundsa 0.052** 0.021 0.069*** 0.075*** 0.086*** 0.040 0.109*** 0.122*** 0.021 -0.005 0.033 0.036 (0.021) (0.032) (0.026) (0.026) (0.031) (0.049) (0.039) (0.039) (0.032) (0.056) (0.039) (0.038) UE fundsa -0.035 0.028 -0.058 -0.056 0.013 0.029 0.019 0.007 -0.064 -0.008 -0.095 -0.081 Feedback effect (0.038) (0.062) (0.047) (0.047) (0.055) (0.096) (0.068) (0.068) (0.054) (0.100) (0.064) (0.063) Sales growthb 0.110*** 0.135* 0.116*** 0.117*** 0.072*** 0.117 0.078*** 0.077*** 0.138*** -0.014 0.161*** 0.167*** (0.020) (0.072) (0.022) (0.022) (0.026) (0.088) (0.027) (0.027) (0.033) (0.123) (0.035) (0.035)

44

(Continuation) Total firms sample Micro and small firms sample Medium and high firms sample Total Pre-crisis Crisis Crisis a. Total Pre-crisis Crisis Crisis a. Total Pre-crisis Crisis Crisis a. Total Firm and environment characteristics Firm size b -0.007 -0.026 -0.002 -0.009 (0.015) (0.024) (0.018) (0.018) High degree b 0.000 0.013 -0.007 -0.007 -0.008 -0.003 -0.012 -0.012 -0.007 0.013 -0.016 -0.014 (0.007) (0.011) (0.008) (0.008) (0.009) (0.015) (0.012) (0.012) (0.010) (0.019) (0.012) (0.012) Tech. park a 0.026 0.175** -0.027 -0.025 0.026 0.169* -0.033 -0.037 0.152** 0.301** 0.112 0.093 (0.042) (0.070) (0.052) (0.052) (0.053) (0.093) (0.066) (0.066) (0.064) (0.122) (0.076) (0.074) Crisis a -0.071*** -0.123*** -0.009 (0.021) (0.030) (0.032) HT manu. a -0.006 -0.073 0.049 0.052 0.055 -0.042 0.128** 0.130** 0.015 -0.035 0.040 0.040 (0.034) (0.050) (0.043) (0.043) (0.044) (0.068) (0.058) (0.058) (0.055) (0.093) (0.068) (0.067) HT serv.a -0.175** -0.250** -0.122 -0.114 -0.094 -0.251** -0.008 -0.015 0.050 0.101 0.069 0.043 (0.070) (0.114) (0.087) (0.087) (0.073) (0.122) (0.092) (0.091) (0.094) (0.191) (0.111) (0.109) MR.1c 0.102 0.053 0.156 0.167* 0.369 0.490 0.321 0.432 0.112 -0.064 0.167 0.183 (0.080) (0.156) (0.097) (0.096) (0.263) (0.424) (0.335) (0.336) (0.113) (0.244) (0.132) (0.130) MR.3c 0.786*** 0.841** 0.767** 0.808** 0.191* 0.183 0.230* 0.234* 0.405 1.002* 0.262 0.341 (0.259) (0.410) (0.330) (0.329) (0.110) (0.218) (0.133) (0.132) (0.289) (0.516) (0.352) (0.349) Constant -0.198 -0.319 -0.323 -0.351 0.420 -0.165 0.425 0.228 0.771* 0.620 0.853 0.669 (0.510) (0.826) (0.657) (0.653) (0.457) (0.779) (0.584) (0.586) (0.456) (0.957) (0.551) (0.546) Sample size 22,573 7,046 15,527 15,527 11,264 3,660 7,604 7,604 11,309 3,386 7,923 7,923 Note: a: Dummy variable; b: Continuous variable in logarithmic; c: Ordered categorical variable. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. See Appendix A for description of the variables.

45

Table 5. Outputs of the growth equation.

Total firms sample Micro and small firms sample Medium and large firms sample

VARIABLES Total Pre-crisis Crisis Crisis a. Total Pre-crisis Crisis Crisis a. Total Pre-crisis Crisis Crisis a.

Inn. Output.b 0.152*** 0.241*** 0.112*** 0.108*** 0.133*** 0.201*** 0.092*** 0.090*** 0.049** 0.075** 0.034 0.047**

(0.017) (0.034) (0.018) (0.019) (0.021) (0.039) (0.024) (0.025) (0.021) (0.032) (0.025) (0.022) Proc. Inn.a -0.003 -0.004 0.008

(0.006) (0.009) (0.008) Non-tech. inn.a -0.002 0.004 0.001

(0.006) (0.009) (0.008) Crisisa -0.114*** -0.128*** -0.100***

(0.005) (0.008) (0.006) Sales Growth (t-1)b 0.965*** 0.913*** 0.974*** 0.975*** 0.972*** 0.939*** 0.979*** 0.979*** 0.973*** 0.919*** 0.981*** 0.978***

(0.006) (0.025) (0.006) (0.006) (0.008) (0.032) (0.008) (0.008) (0.008) (0.028) (0.008) (0.008) Firms size.b 0.025*** 0.035*** 0.022*** 0.023***

(0.002) (0.005) (0.003) (0.003) Export. c 0.014*** -0.010 0.025*** 0.025*** 0.029*** 0.004 0.040*** 0.040*** 0.006 -0.012 0.013*** 0.013***

(0.003) (0.006) (0.003) (0.003) (0.004) (0.009) (0.005) (0.005) (0.004) (0.008) (0.005) (0.005) GITAb 0.004*** 0.003*** 0.005*** 0.005*** 0.007*** 0.004*** 0.008*** 0.008*** 0.003*** 0.002** 0.004*** 0.004***

(0.000) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) Company ageb -0.035*** -0.040*** -0.029*** -0.030*** -0.045*** -0.071*** -0.024** -0.025** -0.018*** -0.012 -0.020** -0.021***

(0.006) (0.010) (0.006) (0.006) (0.009) (0.016) (0.010) (0.010) (0.006) (0.010) (0.008) (0.008) Part. groupa -0.026*** -0.030*** -0.024*** -0.025*** -0.017* -0.025 -0.015 -0.015 -0.014** -0.014 -0.012* -0.014*

(0.005) (0.010) (0.006) (0.006) (0.009) (0.017) (0.010) (0.010) (0.006) (0.012) (0.007) (0.007) PRIVa -0.002 -0.009 -0.001 -0.000 0.029* 0.054 0.023 0.024 -0.006 -0.020 -0.001 -0.001

(0.007) (0.014) (0.007) (0.007) (0.015) (0.033) (0.017) (0.017) (0.007) (0.015) (0.008) (0.008) Tradea 0.008 0.005 0.006 0.006 0.006 0.012 0.002 0.001 -0.001 -0.017 0.002 0.003

(0.010) (0.020) (0.012) (0.012) (0.015) (0.029) (0.018) (0.018) (0.014) (0.028) (0.016) (0.016) Hospitalitya 0.021*** 0.035*** 0.011* 0.011* 0.016* 0.015 0.015 0.015 0.027*** 0.062*** 0.012 0.013

(0.006) (0.011) (0.006) (0.006) (0.009) (0.017) (0.010) (0.010) (0.007) (0.015) (0.009) (0.009) Constructiona -0.034** 0.042 -0.071*** -0.073*** -0.039 0.010 -0.065* -0.068* -0.047** 0.047 -0.084*** -0.083***

46

(0.018) (0.033) (0.021) (0.020) (0.033) (0.059) (0.040) (0.040) (0.019) (0.037) (0.023) (0.022) Financea 0.051** 0.041 0.056** 0.053** 0.104* 0.080 0.112 0.112 0.014 -0.023 0.032 0.030

(0.020) (0.040) (0.023) (0.023) (0.060) (0.104) (0.073) (0.073) (0.020) (0.040) (0.023) (0.023) Constant -0.277*** -0.312** -0.319*** -0.306*** -0.167* -0.140 -0.264*** -0.255*** 0.067 0.239 -0.022 -0.043

(0.065) (0.158) (0.069) (0.069) (0.087) (0.198) (0.093) (0.096) (0.075) (0.168) (0.087) (0.081) Observations 22,573 7,046 15,527 15,527 11,264 3,660 7,604 7,604 11,309 3,386 7,923 7,923 R-squared 0.608 0.334 0.718 0.721 0.638 0.307 0.730 0.731 0.705 0.397 0.772 0.767

Note: a: Dummy variable; b: Continuous variable in logarithmic; c: Ordered categorical variable. GITA: Gross investment in tangible assets; PRIV: Private with foreign participation. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. See Appendix A for further information about the variables

FIGURE

Figure 1. Average firm's sales growth by size.

-15,0%

-10,0%

-5,0%

0,0%

5,0%

10,0%

15,0%

20,0%

25,0%

2006 2007 2008 2009 2010 2011 2012 2013

Micro Small Medium Large